Depending on how you interact with our platform (as a human visitor, a client, or an autonomous AI agent), we may process the following limited data:

• Direct Contact
  When submitting an inquiry through /contact or POST /api/v1/contact, we collect your name, email address, company/organization name, and message payload solely to reply to your inquiry.
• API Key Registrations
  When self-generating an API key via POST /api/v1/keys, optional names or purpose descriptions are retained solely for rate-limit quota allocation and abuse prevention.
• Operational Logs
  For network defense and rate-limiting (e.g. 100 calls/day per network for anonymous MCP calls), request metadata including truncated hashed IP representations and timestamp buckets are processed ephemerally in transient Redis cache.

Our platform provides open Model Context Protocol (MCP) servers, A2A agent cards, and OpenAPI surfaces for AI agents:

• Read-Only Data
  All agent tools (such as get_profile and search_projects) query public, static engineering content. No user session data is injected into external LLM prompts without explicit consent.
• BYOK Sandboxes
  For experimental AI studio features (e.g. Verto AI or a8n integrations), user-supplied API keys are held strictly in memory or client-side storage and never logged in plain text.
• Voice Calls
  Talking to Sara records audio only while you speak. Each utterance is sent to Sarvam AI for transcription, her replies are synthesized there too, and the recording is not stored by this site. Emails and phone numbers you say are replaced with placeholders before the conversation is saved. In Chrome and Edge, the live caption of your own words uses the browser’s built-in speech recognition, which may process that audio on Google’s servers; nothing it produces is saved or sent to the agent.

To deliver global performance, low latency, and dependable email delivery, we rely on trusted infrastructure providers:

• HostingVercel
• CachingUpstash Redis
• Data APIGitHub
• Email RelaySMTP Auth